The Power of the Press: The Effects of Press Frames in Political Campaign News on Media Perceptions

Abstract

This article discusses how the behaviors, roles, and standards of mainstream journalism are framed in political campaign news and empirically examines applicability and accessibility effects of so-called press frames on individuals' media perceptions, specifically on news information quality and negative aspects of content (cf. Kosicki & McLeod, 1990). Following previous work, we posited that journalists frame the press in three ways, called conduit, strategy, and accountability. Participants in a between-subjects experiment were exposed to a political campaign story about an exemplar “character issue” containing these frames. Regarding applicability effects, exposure to each press frame differentially prompted individuals to think frame-relevant propositions were salient in the story. Further, participants exposed to the strategy press frame interpreted the press as being significantly more negative than individuals exposed to the conduit or accountability frame; participants exposed to the accountability press frame interpreted the press as having higher news information quality than individuals exposed to the conduit or strategy frame. Regarding accessibility effects, only the accountability press frame prompted individuals to access prior beliefs about the news media to form opinions about negativity and quality. Implications for future study of press framing are discussed.

Notes

¹Actually, Pan and Kosicki (1993) called their approach “constructivist” and distinguished it in several ways from the constructionist approach of Gamson and Modigliani (1989). These theoretical distinctions pertain more to audience reception than to frame-building.

² Esser and D'Angelo (2003) did a follow-up content analysis of the three press frames after data for this study were collected. They used the term “metacoverage” to refer for this type of news.

³On this point, we followed Iyengar and Kinder (1987) and Iyengar (1991), who argued that participants should be unfamiliar with the specific story topics that they used in their stimuli. Thus, they constructed stimuli to ensure that research participants did not have a great deal of familiarity with the actual stories in their stimuli. Likewise, Cappella and Jamieson (1997; see also Rhee, 1997) drew volunteers from various media markets around the country for their framing experiments. They stated that they wanted to present these research participants with news stories about unfamiliar political candidates (e.g., television news stories about a Philadelphia mayoral candidate being presented to people who live in San Francisco) to control for previous exposure and familiarity with the topic of their stimulus materials.

⁴Discussing the Price–Tewksbury model, McLeod, Kosicki, and McLeod (2002) noted, “Although their knowledge activation model is primarily an organizing model rather than a precise set of hypotheses, it does involve both applicability effects and accessibility effects” (p. 230).

⁵Sample strategy proposition: “Bill Clinton thinks that coverage of the allegations of marital infidelity is creating a crisis in his campaign.” Sample accountability proposition: “Newspaper editors face difficult decisions when figuring out how to report the allegations.” The conduit proposition: “Bill Clinton wants reporters to know that he did not have an affair with Flowers.”

⁶Seven items measured news information quality: (a) “In this story, the news media are trying to report the allegations with a standard of fairness”; (b) “In this story, the news media are giving context on why the allegations are newsworthy”; (c) “In this story, the news media are analyzing their own role in covering the allegations”; (d) “In this story, the news media are trying to cover the allegations in a responsible way”; (e) “In this story, the news media are trying to be impartial to Clinton (neither friend nor foe)”; (f) “In this story, the news media are trying to cover the allegations with restraint”; and (g) “In this story, the news media are trying to act in a professional manner.” Eight items measured negative aspects of content: (a) “In this story, the news media are probing Bill Clinton's personal life in unnecessary detail”; (b) “In this story, the news media are violating the privacy of Clinton and his family”; (c) “In this story, the news media are trivializing the campaign process”; (d) “In this story, the news media are distracting voters' attention away from Clinton's policy positions”; (e) “In this story, the news media do not care if coverage hurts Bill Clinton's chances to win the nomination”; (f) “In this story, the news media are sensationalizing the details of the Flowers allegations”; (g) “In this story, the news media are spreading rumors more than reporting the allegations in an objective way”; and (h) “In this story, the news media are being too negative.”

⁷To explore the orthogonality of these two constructs, factor analysis (principal components analysis, varimax rotation) was conducted on the 15 items composing both constructs. A minimum eigenvalue of 1.0 and at least two loadings (using a 60/40 rule) were the criteria used to retain a factor. A two-factor solution corresponding to the hypothesized constructs was achieved, accounting for roughly 56% of total variance. Each item of negative aspects of content loaded together (eigenvalue = 6.05; 40% of variance explained), except “do not care …” (see footnote 6). Each item of news information quality loaded together (eigenvalue = 2.4; 16% variance explained), except “own role …” (see footnote 5). Items that did not satisfactorily load on each construct were excluded from the indexes. Interitem reliability for each index was high (negative aspects of content, Cronbach's α = .90; news information quality, Cronbach's α = .84).

^a n = 46.

^b n = 47.

^c n = 45.

∗p < .05.

∗∗ p < .01.

^a n = 46.

^b n = 47.

^c n = 45.

†p < .10 (two-tailed).

∗p < 05 (two-tailed).

∗∗ p < .01 (two-tailed).